Am Fam Physician. 2004;70(7):1387
Early detection of bacteremia is clinically important for facilitating rapid intervention before sepsis develops. Jaimes and colleagues report on a prospective study of clinical predictors for bacteremia in hospitalized patients.
This study was conducted at an academic hospital center using 517 inpatients who had laboratory requests for blood cultures. Exclusion criteria included incomplete clinical information, death or hospital discharge within 24 hours, organ transplant within the past year, and pregnancy, all of which removed 17 patients from the study. An extensive list of clinical variables was tracked to determine if any association with positive blood cultures could be established. The variables included age, human immuno-deficiency virus (HIV) infection, chronic renal failure, diabetes, chemotherapy, systemic steroid use, cancer diagnosis, temperature elevation, chills, heart rate, blood pressure, leukocyte count, trauma or surgery, presence of a central venous catheter, mechanical ventilation, length of hospital stay, and use of antibiotics. A positive blood culture was defined as one or more cultures for any recognized pathogen or two or more cultures of a common skin organism, if there also was evidence of a systemic inflammatory response.
More than one half of the inpatients followed in the study had none of the clinical variables. The most commonly identified variables were trauma or surgery, central venous catheter use, chronic renal failure, and diabetes. HIV seropositivity was present in 6.6 percent of patients.
Positive blood cultures, which occurred in 17.8 percent of patients, were associated with a more than 50 percent increase in the mortality rate compared with the overall study cohort. Multivariate analysis identified the statistically significant predictors of bacteremia as age 30 years or older; heart rate 90 beats per minute or higher; temperature 37.8°C (100°F) or higher; leukocyte count, 12,000 cells per mL (12 × 106 per L) or higher; presence of a central venous catheter; and length of hospital stay 10 days or longer. Each of these predictors approximately doubled the risk of a positive blood culture. Excluding blood cultures positive for possible skin flora contaminants (i.e., coagulase-negative Staphylococcus) did not change the results significantly.
The authors conclude that several readily identifiable bedside clinical variables are significant predictors of bacteremia.